Dam-Break Flow Hydrodynamics with Vegetation for DNN Surrogate Modelling

787.1 KB of data from a study investigating deep neural networks as surrogate models for dam-break flows through vegetation. The dataset, authored by Shunsuke Iwasaki and last updated in May 2026, contains results from three-dimensional OpenFOAM simulations validated against laboratory experiments. It was generated to enable rapid assessment of vegetation effects on flooding for nature-based protection strategies.

Use Cases

• Training surrogate deep neural network models for flood hydrodynamics based on simulation data.
• Evaluating vegetation's impact on water surface displacement and horizontal velocity in dam-break scenarios.
• Assessing the extrapolation limits of machine learning models for flow conditions outside a training range.

Strengths

• Data is derived from validated three-dimensional OpenFOAM simulations and laboratory experiments.
• The associated DNN models achieved a mean R² of about 0.80 for the entire time series and 0.93 for the maximum value within the training domain.
• Dataset is openly shared under a CC-BY-4.0 license, facilitating reuse and modification.

Limitations

• Prediction accuracy decreased for vegetation lengths and flow conditions outside the training range, indicating limited extrapolation capability.
• Row count and column-level documentation are unknown, which may limit suitability assessment.
• The dataset is very small at 787.1 KB, suggesting a limited scope of simulation results.

Provenance

Source

figshare

Collection Method

Generated from three-dimensional OpenFOAM simulations validated against laboratory experiments.

Time Range

The study and data generation timeframe are not specified.

Freshness

Last updated 2026-05-08 16:05:04; freshness should be verified.

Geography

Spatial coverage is not specified; likely based on laboratory-scale experiments.